Full text of DOI:10.1002/(SICI)1520-6793(200002)17:2<73::AID-MAR1>3.0.CO;2-L

Th e F u tu r e of
Socia l-Desir a bility Bia s
Resea r ch in Ma r k etin g
Rober t J . F ish er
University of Western Ontario
Social-desirability bias (SDB) research has a long history in the social
sciences in general and in psychology in particular (for a recent review
see Paulhus, 1991). Unfortunately, marketing researchers have not
placed the same emphasis on SDB as their counterparts in other fields.
As King and Blumer point out in the first article in this Special Issue,
only a handful of studies in marketing (primarily in consumer behavior)
actually examine social desirability effects, and less than half of these
report significant findings. Why is this the case? Why haven’t marketers
placed more emphasis on SDB? Why haven’t more significant SDB ef-
fects been found?
Perhaps the most significant barrier to both the testing and evalua-
tion of SDB effects is the complexity of the phenomenon. Evidence sug-
gests that SDB is a multidimensional construct made up of two com-
ponents: self-deception positivity and impression management (e.g.,
Paulhus, 1984). Self-deception positivity is an overly favorable but hon-
est self-presentation that is linked to personality factors such as anxi-
ety, achievement motivation, and self-esteem. Removal of this form of
SDB is not advised because it can actually reduce the validity of content
measures. The impression-management component results from situa-
tional demands or transient motives to present oneself in a positive light
(Paulhus, 1991). Although removing SDB related to impression man-
agement is usually advisable, it may be inappropriate if the component
is conceptually related to the variable of interest (Zerbe & Paulhus,
1987). Moreover, SDB effects are not straightforward because SDB can
attenuate, moderate, or create spurious relationships between variables
(Ganster, Hennessey, & Luthans, 1983).
Psychology & Marketing
᭧ 2000 J ohn Wiley & Sons, Inc.
Vol. 17(2):73– 77 (February 2000)
CCC 0742-6046/00/020073-05
73

Measurement issues provide a further barrier to identifying and min-
imizing SDB. Although a variety of measurement techniques exist,
many are not easily integrated into traditional data collection efforts.
For example, methodologies such as the bogus pipeline (J ones & Sigell,
1971), randomized response technique (Greenberg, Abdula, Simmons,
& Horvitz, 1969), and proxy subjects (Sudman & Bradburn, 1974) go far
beyond pencil-and-paper tests, and therefore add significant cost and
complexity to data collection. As a result, marketing researchers have
avoided these techniques and relied on scales developed in psychology
to reduce or control SDB. In the data-collection stage, marketing re-
searchers have administered an SDB scale along with items measuring
their study’s focal variables (e.g., Richins & Dawson, 1992). Evidence of
scale validity is thought to exist when the SDB scale is uncorrelated
with the target variable(s). Alternatively, researchers have used factor
analysis to identify and eliminate SDB-contaminated items (e.g., Morf
& J ackson, 1972). In the data-analysis stage, SDB scales have been used
as a covariate to partial out SDB effects in regression analysis (e.g.,
Fisher, 1993; Mick, 1996), or to create SDB-corrected measures from
regression residuals (e.g., Fisher & Tellis, 1997).
Despite the relative ease with which SDB scales can be integrated
into questionnaire-based designs, three characteristics of popular SDB
scales appear to limit their application in marketing.
1. Current SDB scales are simply too long and unwieldy to admin-
ister. The venerable Marlowe – Crowne SD scale is a case in point.
Most researchers simply do not have the space to include a 33-
item SD measure on their survey, and respondents may be reluc-
tant to expend the time and effort to complete such an inventory.
Other scales that are popular in psychology, such as the Edwards
(1957) Social-Desirability Scale (39 items) and the Balaned Inven-
tory of Desirable Responding (BIDR) (40 items), are even longer.
The length of these scales restricts their use, particularly if SDB
is likely to affect only one or two of the variables of interest.
2. Respondents may consider some SDB scale items to be inappro-
priate or even offensive. For example, the Marlowe – Crowne scale
includes items such as, “There have been occasions when I felt like
smashing things,” or “If I could get into a movie without paying
and be sure I was not seen, I would probably do it.” Similarly, the
BIDR includes items that might be distressful to some respondents
such as “I have sometimes doubted my ability as a lover,” and “I
sometimes tell lies if I have to.” Researchers (and research spon-
sors) may be justifiably concerned that including such questions
will lower overall response rates and increase the proportion of
missing data. The concern is particularly important when a non-
student sample is used.
74
FISHER

3. SDB scales are too general to be diagnostic. Typical scale items
reflect fundamental beliefs about honesty, getting along with other
people, manners, and jealousy that are not likely to be strongly
associated with specific consumption behaviors. For example, al-
though strong norms against excessive alcohol use exist in the gen-
eral population, getting drunk is expected or perhaps even pre-
scribed in some situations. Adolescents may believe that it is
appropriate to consume alcohol at a party or before a school dance.
Adults might believe that inebriation is acceptable or even desir-
able on special occasions such as New Year’s Eve and the Fourth
of J uly. These situational and subgroup differences in norms at-
tenuate the relationship between general measures of social de-
sirability and specific attitudes and behaviors. As a result, signif-
icant SDB effects are underestimated or overlooked
Accordingly, the lack of SDB research in marketing is perhaps un-
derstandable given the length, sensitivity, and general nature of SDB
scales. The development of scales that are relatively short, inoffensive,
and specific to the targeted attitude or behavior is needed to address
this issue. In addition, it is critical that these scales can be easily in-
corporated into the same data collection effort as the variables of inter-
est because of the variety of situational and research design factors that
influence SDB levels (e.g., research anonymity, respondent instructions,
and sample characteristics). Ideally, the development of new SDB scales
will spark renewed interest in the area, and SDB testing will become a
routine part of the marketing research process. Perhaps a richer un-
derstanding of SDB, and the conditions under which it should be con-
trolled, will also result.
Role of th e Sp ecia l Issu e
The Special Issue reflects a continued interest in SDB in terms of both
our understanding of the phenomenon and the development of new
methods to control it. The Special Issue begins with a critical review of
SDB research in marketing by King and Blumer. The article provides a
useful foundation for thinking about research designs that foster SDB,
the various methods that are typical in the field, and the findings that
researchers can expect. The article highlights the current reliance in
marketing on SDB scales developed in psychology, and the general in-
ability of marketing researchers to find significant SDB effects.
In the second article in this issue, Fisher and Katz use the Marlowe –
Crowne SD scale to estimate the SDB component in self-reported val-
ues. Rather than the traditional conceptualization of SDB as a trait-
irrelevant variance, the article argues that self-reported values should
contain a significant SDB component because values are a conception
SOCIAL-DESIRABILITY BIAS RESEARCH
75

of what is culturally desirable. The research finds substantial differ-
ences in value rankings depending on whether SDB is controlled or not.
An important implication of the research is that raw self-report mea-
sures and those corrected for SDB provide different insights into private
(i.e., anonymous) versus public behaviors within a culture.
The article by Louie and Obermiller examines the effects of SDB on
charitable donations. The authors found that although subjects evalu-
ated the charity director’s gender and the occupation of his or her spouse
to be irrelevant in their donation decision, the factors had significant
effects on financial contributions. It appears that subjects were unwill-
ing to acknowledge the impact of negative gender stereotypes on their
donations because of SDB. The research illustrates that between-sub-
jects experimental designs can mitigate the effects of social-desirability
bias and offers a peer interpretation technique that can be used to in-
terpret these effects a posteriori.
J o describes a methods-factor approach to control social-desirability
bias in structural equation models, and proposes that this approach can
be used whenever multimethod techniques are used. J o presents a how-
to approach for creating methods factors with the use of direct (i.e., first-
person wordings) and indirect (i.e., third-person or projective wordings)
questioning methods across a variety of structural models. By following
the steps outlined by J o, researchers will be able to both identify the
degree to which SDB is significant, and control for its effects in a variety
of causal modeling situations.
Finally, the article by Middleton and J ones reveals the dangers of
blindly applying SDB measures across cultures. The results indicate
that the reliability of the Marlowe – Crowne SD scale is significantly
lower when used in Eastern as opposed to Western cultures. Particu-
larly interesting is the wide variance that the authors find in the item-
to-total correlations of specific items across samples, suggesting that the
items have culture-specific meanings. The article also highlights the
greater importance of social desirability in Eastern cultures and the
need for scales that are tailored to the culture within which they are
employed.
CONCLUSION
The future of SDB research depends upon the discipline’s collective abil-
ity to develop better methods of measuring and ultimately controlling
this form of bias. It is a safe bet that researchers in marketing and other
social sciences will continue to use traditional SDB measures for scale
development and hypothesis testing. Nevertheless, the capacity to iden-
tify and evaluate SDB effects should improve dramatically as new mea-
surement approaches are developed, as should our understanding of the
conditions under which SDB effects occur.
76
FISHER

In closing, I would like to thank all of those who were involved in
making this Special Issue possible. Thanks to Ronald J . Cohen, Editor-
in-Chief, and Rajan Nataraajan, Executive Editor, for giving me the
opportunity to be the Special Issue editor and for their guidance and
support. Thanks to the reviewers for their critical but constructive input
into the review process, and of course, to the authors who contributed
their manuscripts. I hope that you, the reader, find this special issue
both interesting and informative.
REF ERENCES
Edwards, A. L. (1957). The social desirability variable in personality assess-
ment and research. New York: Dryden Press.
Fisher, R. J . (1993). Socially desirable responding and the validity of indirect
questioning. J ournal of Consumer Research, 20, 303 – 315.
Fisher, R. J ., & Tellis, G. (1997). The use of indirect questioning to overcome
social desirability bias: Is the cure worse than the disease? In J . W. Alba &
J . W. Hutchinson (Eds.), Advances in consumer research (Vol. 25, pp. 563 –
567). Provo, UT: Association of Consumer Research.
Ganster, D.C., Hennessey, H. W., & Luthans, F. (1983). Social desirability re-
sponse effects: Three alternative models. Academy of Management J ournal,
26, 213 – 331.
Greenberg, B. C., Abdula, A. L., Simmons, W. L., & Horvitz, D. G. (1969). The
unrelated question in randomized response model, theoretical framework.
J ournal of the American Statistical Association, 64, 520 – 539.
J ones, E. E., & Sigall, H. (1971). The bogus pipeline: A new paradigm for mea-
suring affect and attitude. Psychological Bulletin, 76, 349 – 364.
Mick, D. G. (1996). Are studies of dark side variables confounded by socially
desirable responding? The case of materialism. J ournal of Consumer Re-
search, 23, 106 – 119.
Morf, M. E., & J ackson, D. N. (1972). An analysis of two response styles: True
responding and item endorsement. Educational and Psychological Measure-
ment, 32, 329 – 353.
Paulhus, D. L. (1984). Two component models of socially desirable responding.
J ournal of Personality and Social Psychology, 46, 598 – 609.
Paulhus, D. L. (1991). Measurement and control of response bias. In J . P. Rob-
inson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measures of personality and
social psychological attitudes (pp. 17 – 59). San Diego: Academic Press.
Richins, M. L., & Dawson, S. (1992). A consumer values orientation for mate-
rialism and its measurement: Scale development and validation. J ournal of
Consumer Research, 19, 303 – 316.
Sudman, S., & Bradburn, N. M. (1974). Response effects in surveys. Chicago:
Aldine.
Zerbe, W. J ., & Paulhus, & D. L. (1987). Socially desirable responding in or-
ganizational behavior: A reconception. J ournal of Management Review, 12,
250 – 264.
SOCIAL-DESIRABILITY BIAS RESEARCH
77